I have been asked to review this document on behalf of the OPS DIR.  Overall, the document is well-written and clear.  It provides reasoning for the pagination feature, details error handling, and provides good examples.  It is very much on the right track.

I have two "bigger" questions/comments, and then some nits.  The NETCONF version of this is a normative reference, which is good.  But I wonder if something should be stated about cursor support not requiring state on the server.  This is stated in the NETCONF pagination draft, but I think it might short-circuit worry about extra state if mentioned here, too.

Second, I think this document could benefit from text that cautions that complex pagination (especially sorting and locale parameters) might put performance strain on constrained devices.

Now, onto nits.

Section 2.2

You appear to have an errant double-quote in 'the XML encoding", as required by Section 4.3 of'.  That double-quote should probably be removed.

Section 2.3

s/sublist-list/sublist-limit/

Section 2.3.1

s/unsigned 32 bit integer/unsigned 32-bit integer/

Section 2.3.2

s/should the skipped over/should be skipped over/

You say, "These six query parameters correspond", but there are eight parameters.  This feels like text leftover from an earlier draft and should be bumped up to "eight" to match the list count just above this paragraph.

Sections 2.3.2 and 2.3.3

s/as error-app-tag value/as the error-app-tag value/

Section 2.3.3

s/then a a "501 Not Implemented"/then a "501 Not Implemented"/

Section 2.3.6

s/identity in as the error-app-tag value/identity as the error-app-tag value/

s/MUST be return with the error-type/MUST be returned with the error-type/

Section 3.1

s/All the registrations are to use this document/All registrations are to use this document/